A Tilting Membership
Labour Conference is currently underway. Constituency Labour Party groups (CLPs, or local organisations at the parliamentary constituency level) fund ‘footsoldier’ members to attend the annual meeting, alongside ordinary members attending independently, trade union and affiliated organisation delegates, and the political, policy and journalism professionals for whom attendance is part of their job. Unofficial fringe events also take place concurrently, notably The World Transformed.
This year Conference has coincided with another crisis within the broader leadership, with a motion brought to the ruling National Executive Committee to abolish the Deputy Leader role close to being passed, and the head of policy announcing resignation due to life-work balance issues. It’s therefore a useful point to examine the changing membership in terms of their attitude to the party, the leadership, and on a range of political attitudes and values.
The changing membership has been dealt with insightfully and thoroughly by the Party Membership Project team, in their highly-readable and very new book on political party members, and in a range of papers. I particularly recommend this on the nature of the Labour membership change: Whiteley, P., Poletti, M., Webb, P. and Bale, T. (2019), ‘Oh Jeremy Corbyn! Why Did Party Membership Soar After the 2015 General Election?’ British Journal of Politics and International Relations.
The PMP team gathered bespoke data on party members via a range of surveys fielded between 2015 and late 2018 which will eventually be archived online. Pending those releases, we can extract further detail by an alternative source, via the British Election Study Internet Panel. The BESIP began fielding a question on whether respondents were a member of a political party in Wave 6, and again in waves 8, 10, 11, 12, 14 and 15.
When using the BESIP as a source, we should bear in mind that we don’t know how representative responses are of the entire membership — they are likely to be far more digitally-engaged, for example. Accordingly, we should interpret data as representing Labour members in the BESIP rather than the membership more broadly (while keeping an eye out for data which could allow us to calculate rim weights). We should also bear in mind that ordinary party members only overlapped partially with the selectorate for the 2015 and 2016 leadership contests. However, the size of the BESIP sample also means that we have reasonable sample sizes for Labour members on each wave:
Wave 13 looks anomalous, perhaps because the question was pulled early during the fieldwork, and accordingly I set it aside. Even so, it is tricky to examine the same individuals’ views over time, since very few people respond to every wave, and party membership itself exhibits considerable churn. Just 14 individuals were consistent party members and responded to each of waves 6, 8 , 10, 11, 12, 14 and 15.
Accordingly, here I compare the set of people who reported membership in wave 6 (May 2015) and those who reported membership in wave 15 (March 2019).
Besides responding only intermittently to the BESIP, many individuals dip in and out of membership, presumably for a number of reasons besides an explicit decision to leave or join: they forget to renew or cancel subscriptions, or may forget their precise status at any one point in time. Those cancelling a direct debit remain on the register for some months, and students pay annually, for example. Besides ideological fit, membership decisions also relate to life-stage and a broader desire to be active in associations and community life, and so we would expect the membership to evolve over time even if the leadership and the party’s main policy positions remained fixed.
Only 106 of those responding on both waves (n = 11,968) report membership on both wave 6 and wave 15. Of those who reported membership on wave 6 responding on wave 15 (n=113), 94 percent were still members. Of those who reported Labour membership on wave 15 who had also reported on wave 6 (n = 123), 86 percent also reported membership at wave 6. These rates suggest greater membership retention than an earlier complete case analysis implied, and doesn’t feel plausible; it perhaps reflects conscientiousness associated with both membership stickiness and survey retention.
Demographics of both samples
Bearing in mind ‘normal’ churn, panel attrition and the extent of overlap, the demographics of members in the two waves compare as follows.
Taking the remaining sample at face value, the March 2019 sample of members does look less heavily male-dominated. It also looks considerably younger, with about a quarter of the May 2015 sample born from 1980 but about half of the March 2019 sample.
There is notable missingness for ethnicity and occupational status in Wave 15, at least in the version of the BESIP dataset currently available; but from the data we have, there is no clear change in ethnic composition or in dominance by the middle classes.
Educational status is also of interest, and here we use the most recent data available if missing on Wave 15. For both the sample of members at Wave 6 and Wave 15, 46 percent have an undergraduate degree as the highest qualification achieved, and one-in-five a postgraduate qualification such as a Masters degree.
There is a noticeable difference in the percentage who do not choose a religious affiliation between the two samples, presumably due to the age composition having changed. However, the percentage attending a place of worship at least monthly remains the same, at 15 percent. There is a known association between religious practice and civic engagement; given the shift in balance of membership composition towards the unreligious, it seems likely that those staying in the party and/or those joining the party are disproportionately made up of relatively frequent attenders.
Political Attitudes and Values
We can also examine political values and attitudes to compare members responding at Wave 6 and Wave 15. Ideally we would have a full set of responses for those indicating membership status at both time points. One difficulty (again) is that the more conscientious are likely to be consistent in responding to multi-wave surveys, but this is in turn correlated with authoritarianism, so that the responses we observe from those who respond consistently tend to be from those who are less liberal. We nevertheless report a range of attitudes and values measures while noting that the more liberal (and perhaps the more left-wing) are less likely to have responded at both time-points. Accordingly, it may be more useful to compare the measures outlined in red. However, if comparing May 2015 members with May 2019 members in either May 2015 or May 2019 (rather than May 2015 members in May 2015 with March 2019 members in March 2019), readers should pay attention to the sample sizes reported beneath each measure.
We do see that among those responding in May 2015, there was essentially no difference between the Wave 6 membership sample and the Wave 15 membership sample in their identification as left-wing or right-wing on a 0–10 scale. By March 2019, the Wave 6 sample who responded again feature a slightly higher score, while the Wave 15 membership sample identifies similarly to the subset who responded in May 2015. To reiterate, though, it is likely that the more liberal-left Wave 6 member respondents were simply less likely to respond in March 2019.
It is also likely that individuals’ identification as left-wing or right-wing is not wholly manifested in their economic policy preferences. Accordingly, the British Election Study asks a range of questions to have a more reliable composite measure of leftist or rightist policy preferences, namely extent of agreement that:
Government should redistribute income from the better off to those who are less well off
Big business takes advantage of ordinary people
Ordinary working people do not get their fair share of the nation’s wealth
There is one law for the rich and one for the poor
Management will always try to get the better of employees if it gets the chance
Respondents are scored zero if they completely agree with all five, and 10 if they completely disagree. On this basis, there is no clear evidence that the Wave 15 membership sample is more left-wing, unless we take the Wave 6 membership sample responding in March 2019 as a comparator (and again there is marked attrition between the two time points).
There is a similar approach to assessing liberalism versus authoritarianism, with respondents asked how far they agree:
Young people today don’t have enough respect for traditional British values
For some crimes, the death penalty is the most appropriate sentence
Schools should teach children to obey authority
Censorship of films and magazines is necessary to uphold moral standards
People who break the law should be given stiffer sentences
Respondents were scored zero if they disagreed completely with each statement, and 10 if they demonstrated complete agreement with all 5. The wave 6 membership sample scored 4.3 on average in May 2015, and the wave 15 sample 3.9 in March 2019 (though the minority of Wave 15 sample members responding in May 2015 appeared rather more authoritarian).
If looking at immigration attitudes, there is perhaps some evidence that the Wave 15 sample is a little more liberal than the Wave 6 sample, whether looking at the subsample responding in Wave 6, or when comparing attitudes of the Wave 6 membership sample in May 2015 with the Wave 15 sample in March 2019.
However, we see little clear difference between the membership samples in their chosen position on the EU integration scale, where 0 indicates a preference for full unification with the European Union, and 10 for ‘preserving our independence’. The Wave 6 membership sample scored 2.5 on average in May 2015, and the Wave 15 sample 2.3 in March 2019.
These differences (or lack of differences) are interesting because the folk wisdom is that the post-Corbyn members have distinct political values compared with the pre-Corbyn membership. There is a little evidence for there being differences on the libertarian-authoritarian dimension but very little for the economic left-right dimension. The change in generational composition may account for this.
Demographics and values are not the only relevant areas of similarity or difference, however. In the final sections, I turn to a range of measures of non-electoral political participation, and differences in sentiment towards or liking for the party leader and the party itself.
Differences in Political Activism
A range of measures relating to non-electoral participation were included on Wave 6 of the BESIP, which we could potentially use to examine differences in the nature of political involvement between the Wave 6 membership sample and Wave 15 membership sample.
It is again complicated by wave non-response and panel attrition, and it is very plausible that the Wave 15 membership sample responding at Wave 6 (about a third of the Wave 15 membership sample) are atypical in various respects. Nevertheless I provide a comparison here for completeness, specifically the rates at which members at both time points had ever:
- done any work for a political group by Wave 6
- contacted a politician or government official
- taken part in a demonstration
- signed a petition on the internet
- engaged in a boycott or buycott for ethical reasons.
To the extent that the 270 Wave 15 Labour members who responded at Wave 6 reflect the Wave 15 membership, the differences between the pre-Corbyn and post-Corbyn membership samples look quite small in most cases — contacting officials or politicians, signing petitions, taking part in demonstrations and so on.
The clearest difference is in having done any ‘work’ for a political organisation before Wave 6. Anecdotal accounts suggest that new members have been rapidly socialised into canvassing and leaflet-dropping, and so it is likely that if the question had been repeated at Wave 15 the difference would have been rather smaller.
Leader and Party Liking
Where we can perhaps see more interesting differences between the two samples is in their identification with the Labour Party, and with the party leader. While we only have party membership data for a selection of the later waves, we do have measures for how much respondents like or dislike each major political party, and party leader, on 0–10 scales for each of the 15 waves.
In the two graphs below, I plot the scores for the Wave 6 membership sample responding at each wave, the Wave 15 membership sample, those who have never been a Labour member (effectively the full BESIP sample at each wave), and those who report that they would vote Labour were a General Election held tomorrow.
(Note that these are not the same sets of individuals responding at each and every point, but the samples of those who indicated membership at Wave 6, and those indicating membership at Wave 15, who did respond at each separate wave.)
Between waves 1 and 6, the Wave 6 member samples responding at each wave give higher scores on average to Ed Miliband than the Wave 15 sample. From Wave 7 onwards, the Wave 15 member samples give higher scores, which increase between waves 7 and 9, dip after the Referendum result, and then rise again during the 2017 General Election campaign and following its result. They then dip again for waves 14 and 15.
We see similar patterns for party liking. The scores for the Wave 6 members begin high and generally remain so, though there is a large decrease of two points between Wave 13 (June 2017) and Wave 15 (March 2019). For the Wave 15 members the series begins at 6.5 and generally climbs, although appears to dip in the final two waves.
I also provide leader and liking series for those who report that they would vote Labour in a General Election tomorrow, and the full sample (excluding those indicating Labour membership at any point) for comparison, with the relevant wave sample weights applied in these cases.
On average, the non-member respondents demonstrate fairly consistent dislike for both party and leader, notable given the relatively strong performance by Corbyn and for Labour in June 2017. Among those who would vote Labour at a General Election were it held tomorrow, leader and party liking is consistently strong, not least because those who dislike the party are unlikely to intend to vote for it (and again, this is not the same group of people over time, but those who indicated putative electoral support for the party at each wave). But even so, among those who support the party strongly, there is evidence of declining leader affect from June 2017.
Conclusion
The Wave 15 membership sample is relatively small (N = 777), the demographic data is not complete in every case, and it is very likely that this sample is less enthusiastic on average than, for example, the sample captured at the height of the Corbyn surge at the point of the Smith-Corbyn contest.
Nevertheless, the comparison is useful because we are able to pick up data on just over 1000 members of the pre-Corbyn party, for a comparison with the most recent microdata publicly-available, gathered earlier this year.
The March 2019 membership sample does look quite different from the May 2015 membership sample in age composition; gender balance has improved in the March 2019 sample; both samples look relatively highly-educated and middle-class; and both identify as (and are identifiably) left-wing. On this last, it is not clear that there are large differences in left-right position. It is plausible that the Wave 15 sample is a little more ‘small-l’ liberal, and it is perhaps marginally more pro-EU.
Clearer differences between the two membership samples lie in their attitudes towards and liking of the party and in particular the party leader.
Even so, despite the stronger tendency to like Corbyn as leader — with extremely high liking scores on display — Wave 15 members nevertheless appear to respond to electoral results, notably the EU Referendum result, and the long period of political uncertainty following the 2017 General Election. Those who were members at Wave 6 show, on average, markedly lower liking by March 2019. How far such members and attitudes characterise the current selectorate remains to be seen, whether via a fresh sweep of the YouGov panel, or indeed a future contest.
