Cabinet reshuffles during the Thatcher era

I wanted to do a quick blog about Cabinet changes during Margaret Thatcher’s prime ministership, because we are hypothesizing (in our diaries project) that substantial reshuffles would have immediately impacted the structure of her ministerial meetings.

In a 2005 study it was found that, on average, Cabinet changes in Westminster Parliamentary systems take place once every 11 months1. During her 11.5 years as prime minister, Thatcher conducted mid-term changes an average of once every 7 months.

A brief summary of Thatcher’s mid-term reshuffles is shown in the table below. It is notable that the rate of changes accelerated during her premiership, with four in her first term (one every 12.5 months), and nine in her third term (one every 4.5 months).

TermTotal reshufflesMinisters addedMinisters removedMinisters moved depts
1979 to 198346617
1983 to 198768712
1987 to 19909121316

Despite increased changes over time, the scale of reshuffles in her first and second terms are actually quite similar. In the former she made a total of 29 changes (ministers added + ministers removed + ministers moved to a new department). In the latter she made 27.

Thatcher’s third and final term was the real outlier, however, with a total of 41 mid-term changes. This is a considerable turnover, bearing in mind there are only 21-23 senior ministerial posts in the government. It is also not surprising, based on this information, that Thatcher was isolated and lacked Cabinet support when she was finally usurped by her own party in November 1990.

I should also emphasise that I am only describing reshuffles outside of elections. In addition to the figures mentioned above, a total of 16 changes were made following the 1983 General Election, and a further 18 after the 1987 election. In total, after she entered Downing St in May 1979, Thatcher made 131 personnel changes within her Cabinet.

  1. Kam, C., & Indridason, I.H. (2005). The timing of cabinet reshuffles in five Westminster parliamentary systems’, Legislative Studies Quarterly, 30, 327-363.

Understanding the British core executive

A core aim of our diary research is to provide new knowledge on the diarists we are studying. So far, our focus has mostly been on Margaret Thatcher and her interactions with cabinet ministers , but we also have access to data for Harold Wilson and various ministers/officials at the Department of Trade and Industry (1987-93).

On a wider level, though, our work will inform ongoing debates in the political sciences, and possibly other fields such as contemporary British history and elite studies. One of these debates relates to the nature and workings of the British core executive, or ‘the Centre’, which lays at the heart of democracy in the UK. I believe that our research on Prime Ministers will particularly illuminate the processes of policy formation in Whitehall.

The Centre in the UK can be thought of as a collection of institutions and networks in which the Prime Minister, whilst not endowed with presidential-levels of power, can certainly be considered ‘first among less equals’1. The prerogatives conferred to the PM, such as the ability to appoint (and dismiss) ministers, choose cabinet committees, and conduct bilateral negotiations with department heads, gives them unparalleled oversight of the government’s policy agenda.

In terms of how the PM exercises power depends greatly on the individual.  It has been suggested, for example, that the PM’s office can be stretched like an elastic band to accommodate an assertive individual, or contracted for someone with a more relaxed approach2. The extent of their power at any given time, however, is contingent on a host of mutable factors such as their own personal conduct, public popularity, party support, and level of policy success.  “A week is a long time in politics”, Harold Wilson famously said.

It’s also accepted that the Centre has gradually evolved over time. The PM’s office  expanded under New Labour, for example, when attempts were made to systematically reorganise the Centre into something more coordinated and performance driven3. Today, under the auspices of Boris Johnson, further centralisation of power is afoot. Johnson’s chief adviser Dominic Cummings has been given carte blanche to shake up Whitehall, citing “profound problems at the core of how the British state makes decisions”, and vowing to make  “big changes” during the government’s honeymoon period4.

Despite all of the above, the role and undertakings of the PM are still shrouded in mystery and convention. Our network analytic approach will help to demystify it to a certain degree, and make it possible to compare the actions of core executive actors under different administrations.  This is a important because a functioning democracy requires its citizens to have a better understanding of the decision-making process, and needs checks and balances “to guard against the abuse of power by one or a few powerful policymakers”5.


[1] Heffernan, R. (2003), “Prime ministerial predominance? Core executive politics in the UK”, The British Journal of Politics and International Relations, 5, 347-372.

[2] Jones (1990) Mrs Thatcher and the power of the PM, Contemporary British History, 3:4, 2-6.

[3] Burch, M., & Holliday, I. (2004), “The Blair government and the core executive”, Government and Opposition, 39, 1-21.

[4] Cummings, D. (2020, January) ‘Two hands are a lot’ — we’re hiring data scientists, project managers, policy experts, assorted weirdos…. Dominic Cummings’s Blog.

[5] Dunleavy, P. (2018), “The core executive and government”,  In P. Dunleavy, A. Park, Taylor, R. (Eds.). The UK’s Changing Democracy: The 2018 Democratic Audit. London: LSE Press.









Heathites & Thatcherites as covariates

An interesting aspect of Margaret Thatcher’s early cabinets is that they comprised a number of Tory grandees who were hostile to her domestic agenda. These political foes were mostly Heathites (i.e. they were loyal to former party leader Ted Heath), alternatively known by Thatcher’s supporters, rather disparagingly, as ‘Wets’.

The Wets favoured a politics of consensus, and were not averse to government intervention in the economy to sustain full-employment. Such a ‘paternalist’ instinct was anathema to Thatcher. She was a self-styled conviction politician, and firmly believed that publicly owned industries should be competitive in the free market, without recourse to government bailouts.

Despite her stated convictions, Thatcher did not immediately purge her shadow cabinet of Heathites when she entered Downing Street, nor did she promote a host of Thatcherites (or ‘Dries’) to senior ministerial positions. Her supporters were relatively junior members of the party compared to Heathites, and the composition of her first cabinet (9 Wets and 4 Dries) was tacit acknowledgement of this fact. In other words, she tread carefully to prevent a revolt from the backbenches.

The distinction between parliamentary Wets and Dries is useful for our appointment diaries research. We’ve included a ‘Dryness’ effect in our models to compare findings relating to both factions. We have been investigating, for example, whether (and how) Thatcher used meeting structure/composition to drive through her agenda. Hopefully I can share the results of our RHEM modelling shortly. In the meantime I wanted to comment on some interesting preliminary findings, which are based on a cross-sectional one-mode analysis of the PM’s ministerial network.

The line graph below shows the ‘betweenness centrality‘ scores of  both Wets and Dries throughout Thatcher’s first term. It’s notable that, as time progressed, Thatcherites increasingly bridged groups of ministers meeting the PM, whereas Heathites did so much less frequently, and on a diminishing basis.

Wets dries

The findings tell us at least two things:

  1. ‘Wets and Dries’ is not merely an arbitrary concept, and as such, it merits further investigation.
  2. Thatcher’s supporters were well placed to (theoretically, at least) control the flow of information across the ministerial network.

The second point is the most important one. Based on this analysis it could be hypothesised that the PM used meetings to implement divide an rule tactics during her first term. This is a persuasive theory considering internal opposition to her policies, and is perhaps something we can substantiate with more analysis.

Thatcher, Johnson, and crisis meetings

I’m in quite a curious position right now. My research project, which examines political networks through the lens of appointment diaries, has suddenly become very topical.

For the best part of a year I have been poring over Margaret Thatcher’s meeting schedule, becoming increasingly acquainted with the inner workings of the British core executive – and now the PM’s day-to-day activities have been projected quite vividly (luridly, even) into our collective conscience. I feel a bit cold about it all, though, for reasons that are probably obvious.

We learned this week that Boris Johnson missed five Cobra meetings on the Covid-19 pandemic. These are briefings where the British government co-ordinates its response to national or regional crises. Johnson’s non-attendance has led to widespread criticism in the media, which is understandable given the UK’s alarming (and frankly depressing) rate of Covid fatalities compared to most other nations globally. He has, however, been defended by cabinet minister Michael Gove, who highlighted that PMs do not always attend Cobra meetings, and so his absence was not out of the ordinary.

Reading about the Cobra fiasco made me think about some of the analysis of Thatcher’s diaries that I’ve done so far. There is a particularly interesting period during the 1979 Parliament when her ministerial network altered markedly, and it was during the Falklands Crisis. I thought that I’d share some of these findings, because they are quite evocative of the current situation in Whitehall.

The simplest (and most striking) way to present the results is via network diagrams. First, let’s consider Thatcher’s meetings with cabinet ministers in March 1982 – the month immediately prior to the Falklands conflict.

35 Mar-82

This is fairly typical of the PM’s ministerial network at that time. When Thatcher entered office in May 1979, she faced a lot of internal opposition from senior members of the Conservative Party, who considered her to be an untried extremist. By September 1981, however, she had shifted the balance of power in her favour by promoting several of her supporters to the cabinet, and removing some of her opponents. As a result, the overall density of her networks increased notably in subsequent months, and fewer ministers were left on the sidelines.

Now let’s consider April 1982.

36 Apr-82

We see here that the Defence Secretary, the Foreign Secretary, the Home Secretary, and the Chancellor of the Duchy of Lancaster were suddenly the intense focus of the PM’s attention. These ministers were all on the hastily arranged Falklands War Cabinet, which convened 66 times between April and July 1982, when the conflict ended. The PM continued to meet all of her cabinet ministers, but the thinning of the edges (lines between nodes) shows that, for the most part, these appointments were less frequent than they were in March.

Finally, a look at May.

37 May-82

Meetings with the War Cabinet continued apace, as would be expected. However, in the thick of the crisis, the PM stopped meeting many cabinet colleagues altogether. It is interesting that this didn’t happen sooner – Argentina invaded the Falklands on 2nd April. Perhaps it’s because the British Task Force took weeks to reach the South Atlantic. Or maybe it reflects Thatcher’s general reluctance to refuse meetings.

From the analysis I have done so far, no other period of Thatcher’s premiership resulted in such a profound network contraction. I imagine that, having returned to work after a near-fatal bout of Coronavirus, Johnson’s ministerial network will look not entirely dissimilar to this, as the SAGE Committee attempts to navigate us through the pandemic.

Ultimately for Thatcher, the Falklands War, and her decisive (albeit reactive) response, sealed her second term in office. It remains to be seen, however, what effect the current crisis will have on Johnson premiership.

* Network diagrams omit meetings of the full cabinet, and are instead based on cabinet committees and ad hoc meetings with ministers.

Studying Thatcher’s diaries as multi-actor events: introducing the relational hyperevents model

Our project team is engaged in an ongoing collaboration with Jürgen Lerner of the University of Konstanz, Germany. Jürgen specialises in statistical network modelling, and has been working with us to develop models that can be fitted to appointment diary data.

So far, this work has resulted in a draft paper in which Relational Hyperevent Models (RHEM) are proposed as a means of studying multi-actor networks. It is predominantly a methodological paper, and therefore some of it is quite technical. However, it uses two illustrative examples – one on Margaret Thatcher’s ministry,  another co-author networks –  to show how the underlying structure of such networks can be studied using advanced modelling techniques.

The RHEM framework is a generalisation of the Relational Events Model (REM). REMs make it possible to study network effects from chronological and time-stamped event data. Applications of REMs have mostly focused on dyadic network data i.e. events which take place between pairs of actors in a bounded network. For example, information sharing between people working in teams or birds sharing food in an aviary. In these cases, interactions involve actor a sending an event to actor b, which is followed by a series of subsequent events which may involve other actors in the network.

In our case study, we extract from the Prime Minister’s (PM) appointment diaries time-stamped and chronological lists of interactions with cabinet ministers. However, the PM meets regularly with one, several or all cabinet ministers. Therefore, such multi-actor formations are more appropriately conceptualised as “hyperedges“, as opposed to dyads.

Using RHEMs, we investigate effects such as repetition (i.e. the tendency for events on actors to be repeated) at multiple levels. For example, does the PM tend to repeat meetings on specific hyperedges (i.e. groups of ministers), and/or triads, pairs, or individuals within hyperdges? Crucially, using statistical inference, we can determine whether such effects exist in the PM’s ministerial networks, and if so, whether they are more or less likely than would be expected by chance.

At the moment, we are using RHEMs to develop a substantive case study on Margaret Thatcher’s first term as PM (May 1979 to June 1983). We are focusing on the existence of multi-actor network effects in her engagements with ministers, and:

  1. How they develop or change over time.
  2. How they are impacted by external events such as the Falklands War.
  3. How they are impacted by other factors such as hostile/friendly relationships.

In the meantime I would encourage you to have a look at our draft paper on RHEMs.


A brief network analysis of Thatcher’s first term

The method we are using to study historical appointment diaries is a form of social network analysis that exploits chronological event data. This means we are primarily interested in relational dynamics as opposed to viewing networks as a snapshot in time. Despite this, looking at ‘static’ networks can still be useful to us, and I will demonstrate that here with a brief example.

Our first case study is based on former Prime Minister Margaret Thatcher’s meetings with cabinet ministers. To get an immediate insight into the nature of these exchanges, we can aggregate data for a specific time-period and create a network diagram. By way of example, the first four months of Thatcher’s opening term is shown in Figure 1. (Note: this analysis excludes full cabinet meetings.)

With MT no cabinet
Figure 1: PM meetings with cabinet ministers from May to August 1979

Each node here represents a cabinet minister. Those with larger nodes attended a higher number of meetings with the PM. Relational ties between nodes show that linked individuals attended the same meeting on at least one occasion.  Naturally the PM is linked to all ministers and is represented by the largest node in the network.

Figure 1 also shows that there is a dense sub-group of nodes in the network, and that Geoffrey Howe, Lord Carrington, and Willie Whitelaw are all prominent within this group. This is expected because they occupied the three highest ranking cabinet positions: Chancellor, Foreign Secretary, and Home Secretary, respectively. Perhaps more interesting are the peripheral actors in the network. For example, Nicholas Edwards (Wales) and Mark Carlisle (Education) only have links with the PM, and are both represented by smaller nodes. This means that from May to August 1979 they attended relatively few meetings with the PM, and none involved other ministers (outwith full cabinet meetings).

Figure 2: PM meetings with cabinet ministers from September to December 1979

How did things change in the next four months? As shown in Figure 2, not a great deal – the same dense sub-group exists and most peripheral actors remain on the outside. However, at a glance, two observations can be made:

  • Humphrey Atkins gravitated towards the dense sub-group, and his closeness centrality score increased (Figures 3 and 4). Atkins was the Secretary of State for Northern Ireland at this time, and on August 27th 1979 there were two major incidents involving the IRA which could have been the catalyst for more meetings with the PM and other ministers.
  • Norman St John-Stevas (Duchy of Lancaster) gravitated away from the dense sub-group, and his closeness centrality decreased (Figures 3 and 4). He was removed from the cabinet during the PM’s first reshuffle in January 1981. This could show that he fell out of favour very early during the PM’s first term, long before his eventual sacking.

Blog 1
Figure 3: Closeness centrality per cabinet minister, May to August 1979

Blog 2
Figure 4: Closeness centrality per cabinet minister, September to December 1979

This is a cursory analysis that was conducted mainly for illustrative purposes. Even still, there is clear value in this approach. It: (a) contextualises Margaret Thatcher’s first eight months as PM, (b) allows us to make inferences about the structure of the PM’s formal interactions with ministers, and (c) provides ideas for hypotheses that can be used in the main study, where we will focus explicitly on temporal network dependencies.

The analysis described in this post was conducted on R:

R Core Team (2013). R: A language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria.



Margaret Thatcher diary research

For those of you who don’t know about me or my research interests, please have a look at the updated About page for more details.

It is over two years since I blogged about my work. Time certainly flies. I will spare you the details – suffice to say, I have since completed my PhD, spent a year working for Information Services Division in a statistical governance role (with some user research), and am now back in academia as a research associate at the University of Glasgow.

I am currently working on an ESRC funded project with Professors Neil Rollings (Economic and Business History) and Mark Tranmer (Quantitative Social Sciences). We are studying the historical appointment diaries of politicians and senior civil servants using quantitative methods. You can read a detailed abstract on the project here.


book business calendar close up
Photo by Pixabay on

In essence, the research focuses on three case studies; the first of these is the appointment diaries of former British Prime Minister Margaret Thatcher. Using the Relational Event Model, we are analysing meetings with key contacts (e.g. business representatives, journalists) to uncover significant relational patterns in the data. Still at the formative stage of the work, we are considering the insights we can generate using this approach, such as:

  1. Margaret Thatcher’s leadership style, and how it developed over time.
  2. Event sequences before and after external ‘shocks’ – e.g. the Falklands War.
  3. The nature of interactions between the Prime Minister and key individuals (e.g. special advisers) and/or groups (e.g. the CBI).

At the moment, these aims are quite broad. To narrow them down we are working on a pilot which focuses on meetings between the Prime Minister and her cabinet colleagues between 1979 and 1984 – i.e. when she assumed power until the year after her first re-election.

So why cabinet ministers, and why this time period?

Firstly, it seems like an obvious start from the historical viewpoint. Although I am not an historian, this period heralded significant change in the UK (and international) political, economic, and social context – and Thatcher’s administration was at the vanguard of this change. Notably, though, many in her first cabinet were hostile to her proposed economic policies. As such, studying the evolving dynamics of the relationships between the Prime Minister and her colleagues could yield interesting new insights on her early years at the helm of government.

Secondly from a practical viewpoint, focusing on cabinet ministers provides a clear network boundary to narrow the scope of the research and simplify the coding process. To this end, there are a limited number of ministers at any given time (21 and 22 during the time in question), and no ambiguity about who qualifies as a cabinet minister – it is a matter of public record.

So, this is roughly where we are just now – tentatively moving into the analysis phase. My plan is to use the blog to talk about the project every few weeks, giving updates and commentating on various aspects of the study – method, findings, and anything I think is worth sharing.

If you’d like to get in touch about this work then please see the contact page for details.






Young Scottish people and Twitter

Is there a love affair between young Scottish people and Twitter? My survey analysis suggests that there is. According to the findings, and amongst the respondents who had a Twitter account:

  • 36.6% of those who identify as Scottish used it throughout the day, whilst;
  • 57.9% used it at least a few times a day*

Image credit below**

To put those figures in context, let’s consider the next most prolific users in terms of ethnic background:

  • 20.3% of those who identify as Other British or Irish used it throughout the day, whilst;
  • 46.4% used it at least a few times a day*

It is notable that the other big predictor of frequent Twitter use was age, with 16-18 year olds being much more likely than 19-24 year olds to use it throughout the day. However, when excluding Scottish respondents from the analysis (the largest proportion of whom were in the 16-18 age group), the effect of age on Twitter use, albeit still evident, was considerably less pronounced.

In addition to the above, the Scottish respondents had more Twitter followers, and followed more Twitter accounts than those from different ethnic backgrounds. These findings are interesting, and certainly invite further analysis to elucidate them.

Could it be that there is a distinct subculture of Twitter use amongst young Scottish people? In a separate research project that I have been involved in recently I spoke to many young teenagers in Scotland about their social media use, and would argue that there is evidence of this (and that young people view it as a platform for anarchic, “exciting” content – see this Buzzfeed article for example).

A caveat, though: actual uptake of Twitter is only about half that of Facebook, according to the survey. And although there does seem to be evidence that a faction of young Scottish people do use Twitter heavily, quantitative evidence suggests that they are still a (substantial) minority. A “clique”, even.

*These figures include those who used Twitter throughout the day.

** The Twitter bird © Photo by: Eldh, A. (2011), Weblink here. Licence:

Additional note: The complete survey sample was 909 respondents, of whom 57% were Scottish, 13.6% Other British and Irish, 17.4% Any other White background, and 11.7% Any other background.

Job search stress, gender, and conscientiousness

I’ve been picking up some interesting findings in my survey about job search and social media use. One group of findings in particular jumped out at me yesterday – those associated with job search stress.

As part of the questionnaire, I asked respondents how stressful they found looking for a job. The results were pretty conclusive:

  • 65.2% agreed that it was stressful
  • 14.5% disagreed that is was stressful.


Breaking down these figures, it is clear that the pressure of job search has a similar impact on every demographic subgroup. Indeed, for most demographics, the differences between the variables are negligible. However, one of the findings was eye-catching. With regards to gender and job search:

  • 56.3% of males agreed that it was stressful; whilst,
  • 70.1% of females agreed that is was stressful.

The difference between males and females was statistically significant. Perhaps one of the reasons this finding leapt out at me was that for the bulk of the analysis, gender made little difference one way or the other.

This made me trawl back through my findings to find another area where gender had had a significant impact. An example I found was in respect of job search effort: females (64.2%) were more likely than males (56.5%) to agree that they put effort into job search. Indeed, job search effort transpired to be one of the few dependent variables which significantly impacted job search stress levels:

  • 57.7% of those who did not apply effort agreed that it was stressful; whilst
  • 71.6% of those who did apply job search effort agreed that it was stressful.

In summation, there is a strong association between gender, job search effort, and job search stress*. As an amateur (non) psychologist, these results seem to suggest that females are more conscientious than males when it comes to seeking employment. A cursory Google Scholar search indicates that these findings have some precedent in other contexts.

The obvious question, then, is what impact is this having on job search outcomes**? This is something I will hopefully address in the thesis.

*Males who apply effort to job search (62.5%) are also more likely than those who do not (54.0%) to find job search stressful. Therefore, level of job search effort – independently of gender – also appears to have a significant impact on stress.

** The other obvious question is, what can be done to ameliorate job search stress beyond just treating its symptoms? Is it a wider societal or cultural concern, that people become so stressed out about finding work? Especially young people being under so much pressure to find a foothold in the labour market. However, this could be a whole new PhD project.

Unemployed young people in Scotland: do they actually want a job?

In the UK, cutting welfare benefits  has been used as a tool by the government in recent years to “encourage hundreds of thousands of people into work”. The idea is that reducing state-funded welfare is supposed to drive up wages and therefore, presumably, make employment a more appealing prospect for people out of work. Inherent within this argument, or at least implied by this argument, is the notion that unemployed people in the UK don’t really want to work in the first place.

So what then of 16-24 year olds in Scotland?


Photo © 2011 J. Ronald Lee

I’m currently analysing survey data gathered from 909 young josbeekers based in Scotland. To understand the general nature of each respondent’s job search I included some contextual questions. For example, I asked if they were looking for a job that was a “career option”, which elicited the following response from 2 of the demographic subgroups:

Unemployed respondents: 71.4% agreed that they were looking for a career option*

Higher education students: 36.0% agreed that they were looking for a career option

It’s important to note that in the HE students category, the respondents were at different stages of their qualifications. As such, most of them would have been seeking casual work, instead of graduate or other full-time positions (this is supported by the data). However, when the same sub-groups were asked if they would “settle for any job as long as it pays money”, this was the response:

Unemployed respondents: 61.7% agreed that they would take any job that paid money*

Higher education students: 37.7% agreed that they would take any job that paid money

There is something quite moving about those figures. They show that young people who are unemployed are motivated by the idea of finding a job with actual prospects, but most are prepared to take any job in order to get by. In fact, they are far more likely than students seeking casual jobs to take whatever they can get. And the figures are actually distorted somewhat by unemployed respondents who are educated to university level (see the bottom of this blog entry).

It is also noteworthy that the same unemployed respondents had the most flexible attitude towards working evenings and weekends (90.0% agreed) and were only slightly less amenable to the idea of relocation than those who were employed and also seeking a job.

Within the context of this analysis, it is difficult to afford any credence to the idea that unemployed people are not naturally disposed to the idea of finding a job. To the contrary, the analysis indicates that unemployed young people (a subgroup dominated in this sample by young males with lower educational attainment) are aspirational, yes, but also desperate to find a job of any kind just to have an income. Therefore, shifting the onus to the individual level to explain worklessness, rather than focusing on wider societal or economic conditions, seems to be a rather shady tactic at best.

* The figures were higher for unemployed people with lower levels of education i.e. of those unemployed respondents with no qualifications or National level qualifications (Scottish equivalent of GCSEs) 72.4% agreed that they were looking for a career option (0.0% disagreed), whilst an overwhelming 89.7% agreed they would take any job that paid money (3.4% disagreed).