Measuring Partisanship as a Social Identity in Multi-Party Systems

Instrumental and expressive partisanship are not completely unrelated. Right-left ideology and issue preferences are strongly linked to the direction of partisanship and likely provide an initial impetus to support one party over another. But ideological intensity is only weakly linked to the strength of partisan identity, as we show empirically in this study. Moreover, the stability of partisanship in the face of changing party platforms suggests that issue stances may follow partisanship not vice versa, a proposition at odds with the instrumental model but consistent with numerous studies on elite influence and party cues in the U.S. (Cohen 2003; Dancey and Goren 2010; Druckman et al. 2013; Lenz 2013).

In the following, we largely follow the discussion and notation in De Ayala (2013), pages 209–236.

Roughly 5000 individuals responded to each wave (5195 for the pre, 5225 for post, and 5732 for the values module). For documentation and data, see http://​www.​lissdata.​nl/​lissdata/​. The 9/12 and 12/12-13/1 waves contained additional items needed for this analysis.

The Panel 8 sample is mainly self-recruited (70 %); the remainder 30 % is a probability-based population sample. The opt-in panel was recruited through internet advertising on the websites of newspapers, Twitter, Facebook and blogs. The second sample (Panel 8.2) was drawn using a probability-based recruitment method only.

Panel members are recruited from various sources including advertising on a wide range of website. For each recruited panel member, YouGov records socio-demographic information in order to ensure a nationally representative adult sample in terms of age, gender, and social class. Everyone taking part in a YouGov survey receives a modest cash incentive for their participation. Data can be downloaded at http://​www.​britishelections​tudy.​com/​.

These numbers are based on the entire sample in each country, not just those who received the partisan identity module.

A five-item strength measure (very convinced adherent, convinced adherent, not a convinced adherent, attracted, voted for a party) provides more information than the three-item strength measure on in-party voting in the Netherlands, and its coefficient is comparable in size to that of partisan identity. But it does not account as well as partisan identity for political participation.

Greene used the original IDPG wording which refers to “this group”; we reworded all items to refer to “this party”. Of the 5 items adapted from the IDPG scale, the first three items in Table 2 are taken almost verbatim from the scale. The fourth and fifth items in Table 2 were dramatically reworded. The original item “I have a number of qualities typical of members” was reworded to “I have a lot in common with other supporters of the party” and “If a story in the media criticized this group, I would feel embarrassed” was reworded to “If this party does badly in opinion polls, my day is ruined.” The last three items in Table 2 were newly added to the scale.

Respondents without a party preferences were not asked the partisan identity questions. In Sweden this included those who said they did not feel close to any party, did not want to, or did not answer the question resulting in the omission of 254 respondents. In the Netherlands, 499 respondents were dropped because they were not a supporter nor attracted to a party, had not voted, or left the question blank. In wave 3 of the BES, roughly 13 % (3616) of respondents in the full sample did not provide a party preference. Similarly, 13 % of respondents (2234) in wave 4 of the BES did not indicate a party preference.

We rely on more numerous data from wave 3 for the IRT analysis and wave 4, which contains a political engagement question battery, for the regression analyses.

The Graded Response Model is an extension of the two-parameter logistic (2PL) model for graded response data in the sense that it allows the discrimination parameter α to vary across items.

The theoretical range of theta is −∞ to ∞. However, typical item locations fall within a range of −3 to 3 (De Ayala 2013).

We also calculated the total information provided by each identity item and traditional partisanship strength (Table A2 in the Online Appendix). In each country, the traditional partisan strength item captures less information than any individual partisan identity scale item, and thus provides far less information than the overall partisan identity scale.

Due to similar wording of some items, the error terms for the following items covaried: "When I speak about this party, I usually say ‘we’ instead of ‘they’” and "When I speak about this party, I refer to them as "my party."; "When people criticize this party, it feels like a personal insult" and "When people praise this party, it makes me feel good” as well as "I have a lot in common with other supporters of this party" and "When I meet someone who supports this party, I feel connected with this person." For detailed instructions on how to use the lavaan package, see Hirschfeld and von Brachel (2014).

To further examine whether one or more specific items undermined scalar invariance, we re-ran the invariance analysis and dropped each item in turn. These analyses are included Table A3 (Online Appendix). They suggest that item 8 contributes a little more than other items to weak scalar invariance.

These numbers are obtained from the scalar invariance model for continuous data. Results remain valid if we compare the EPC-interest values to latent mean differences in partisan identity scores obtained from a model for categorical data.

These percentages remain stable across waves 3 and 4.

In the Netherlands, political activities included: (1) making use of a political party or organization, (2) participation in a government-organized public hearing, discussion or citizen participation meeting, (3) contacting a politician or civil servant, or (4) participating in a political discussion or campaign by Internet, e-mail or SMS. In Sweden, activities included (1) contacting a politician, (2) given/raised money to/for a political organization, (3) contacting a civil servant, or (4) attending a political rally. All four questions were repeated in two earlier panel waves, Citizen Panel 5 (11/12/12-12/16/12) Citizen Panel 7 (6/12/13-7/8/13). All 12 items (4 items in 3 waves) were additively combined and rescaled on a 0 to 1 scale. In the U.K., all respondents in wave 4 were asked whether they had visited the website of a candidate or party, signed up or registered online to help a party or candidate, read or found information in the last four weeks tweeted by (1) political parties or candidates, (2) a personal acquaintance, or (3) others, such as a commentator, journalist, or activist, or obtained information from each of these three sources. Five additional questions asked whether the respondent has shared political information through (1) Facebook, (2) Twitter, (3) email, (4) instant messaging or (5) another website or online platform.

Party leaders were evaluated on a scale that ranged from 0 (strongly dislike) to 1 (strongly like) in wave 4 of the BES.

Results in the Netherlands are robust to type of analysis when re-run as an ordered probit model (since the political participation variable has only 5 points).