The study was conducted by Lloyd George Banda of Stellenbosch University and the Albert-Ludwigs-Universität Freiburg, together with Kamal Yakubu of Afrobarometer in Accra, Ghana. Their starting point is expectancy-disconfirmation theory, a framework originally developed in consumer research by Richard Oliver in 1980 and later adapted to the public sector by Gregg van Ryzin. According to this theory, satisfaction is not simply a function of performance; it emerges from a comparison between perceived performance and a prior reference standard called expectations. Performance that exceeds expectations produces satisfaction, performance that falls short produces dissatisfaction, and performance that merely confirms expectations leaves satisfaction moderate. The twist introduced by Banda and Yakubu is to ask what happens when those expectations are political—when the reference point against which citizens judge economic outcomes is structured by the party they identify with.

The Partisan Disconfirmation Model builds on the insight that citizens rarely form expectations about government performance in isolation. Macroeconomic outcomes are complex, delayed, and difficult to attribute clearly to government action, which means citizens confront economic information in environments that are simultaneously information-rich and cognitively demanding. Under such conditions, people rely on heuristic cues, and partisan identity is among the most powerful. Ruling-party identifiers tend to expect favorable performance, opposition identifiers tend to anticipate poorer outcomes, and these expectation differences should, in principle, shape how new economic information is interpreted. The model draws on Hypothesis Testing Theory, which conceptualizes individuals as intuitive hypothesis testers who evaluate incoming information against prior beliefs, often interpreting it in belief-consistent ways due to confirmation bias.

Crucially, the model makes a conditional prediction rather than a uniform one. When economic signals are clearly positive or clearly negative, performance assessments should dominate satisfaction judgments, leaving limited scope for prior expectations to shape evaluations. But when signals are ambiguous—mixed, weak, or uncertain—prior expectations should exert greater influence, conditioning how outcomes are interpreted. In survey terms, neutral evaluations of economic conditions, where citizens report conditions as neither good nor bad, often reflect informational ambiguity rather than a stable assessment. The researchers therefore hypothesized that partisan differences in satisfaction should be strongest when economic evaluations are neutral and attenuated when conditions are clearly favorable or clearly unfavorable. This is the disconfirmation logic: partisanship as a cognitive prior whose influence varies with the clarity of the performance signal.

To test these predictions, the team analyzed nationally representative Afrobarometer survey data from Malawi collected in 2022, covering 1,200 adult citizens with a margin of error of ±3 percent at a 95 percent confidence level. Malawi is a theoretically informative case: a unitary presidential democracy with a pluralistic party system, six transfers of power since 1994, and a vibrant civil society, yet facing significant socioeconomic challenges with an average Human Development Index value of 0.501 and GDP per capita of roughly $574 at constant 2015 prices. The dependent variable was satisfaction with government economic management, measured on a four-point ordered scale from very dissatisfied to very satisfied. The key independent variables were evaluations of current national economic conditions, classified as bad, neutral, or good, and partisan identification, distinguishing opposition supporters, ruling-party supporters, and non-partisans. The analysis estimated ordered logistic regression models with interaction terms to assess whether partisan effects varied across economic contexts, evaluated hypotheses using omnibus Wald tests of joint parameter significance, and conducted post-estimation contrasts based on predicted probabilities.

The descriptive picture alone is sobering. Eighty-five percent of respondents evaluated national economic conditions as bad, and 86 percent were dissatisfied with government economic management. Yet the regression results reveal systematic variation beneath this widespread pessimism. Compared with respondents who evaluated economic conditions as bad, those who evaluated them as neutral were over two and a half times more likely to report satisfaction (odds ratio = 2.69, p = 0.002), while those who evaluated conditions as good were more than three and a half times as likely (odds ratio = 3.67, p < 0.001). The joint Wald test confirmed a strongly significant overall economic effect (χ² = 16.72, p < 0.001), providing clear support for the performance dominance hypothesis. Perceived economic performance emerged as the strongest predictor of satisfaction with government economic management.

Partisan alignment also shaped evaluations, just as the model predicted. Ruling-party identifiers were significantly more likely to report satisfaction than opposition identifiers (odds ratio = 2.401, p = 0.001), and non-partisans fell in between, more satisfied than opposition supporters (odds ratio = 1.363, p < 0.05) but less satisfied than ruling-party identifiers. The joint test of partisanship variables was equally significant (χ² = 19.48, p < 0.001), confirming the partisan alignment effect. Among the descriptive statistics, 46 percent of ruling-party identifiers were satisfied with government economic management, compared with 34 percent of non-partisans and only 20 percent of opposition identifiers. The researchers also checked for endogeneity using a two-stage residual inclusion approach with instruments including religion, presidential performance evaluation, and employment status; all computed residuals were statistically indistinguishable from zero, indicating no significant endogeneity bias in the key variables, though the authors interpret their findings as associations rather than causation given the cross-sectional design.

The surprise came with the interaction tests. Contrary to the disconfirmation logic of the Partisan Disconfirmation Model, none of the nine pairwise interaction combinations between economic conditions and partisan identification were statistically significant. The interaction block test was not significant (χ² = 2.26, p = 0.688), indicating that perceptions of economic conditions did not vary across partisan alignment. Additional tests comparing neutral evaluations with both negative and positive evaluations similarly revealed no systematic differences in the strength of partisan effects, so the moderation hypotheses were not supported. The attenuation hypothesis fared no better: difference-in-differences tests comparing the partisan gap under neutral conditions with the gap under bad conditions (β = −0.794, p = 0.184) and under good conditions (β = −0.832, p = 0.275) found no evidence that partisan differences were larger under ambiguity. In both cases the estimated coefficients were negative, meaning partisan differences were not amplified under neutral evaluations at all.

What explains this absence of conditional partisan belief updating? The authors propose one intriguing interpretation: citizens who hold mixed or uncertain evaluations of economic conditions—those often characterized as cross-pressured, swing, or ambivalent—may not resolve ambiguity by increasing reliance on partisan cues. Rather than exhibiting stronger identity-consistent evaluation under uncertainty, such individuals may process economic and political information in a more fragmented or weakly structured manner. This is consistent with research on cross-pressured voters, which suggests that individuals facing competing evaluative signals may exhibit attenuated or less coherent political responses, and with work on political ambivalence suggesting that competing loyalties can produce cognitive fatigue, indifference, or reduced decisiveness rather than heightened motivated reasoning. The data do not directly measure political disengagement, but the pattern is consistent with forms of ambivalence that future research could examine more directly.

The broader implications are considerable. The findings challenge claims in some African political behavior literature that political attitudes are primarily driven by identity-based loyalties such as ethnicity or partisan attachment. Instead, the evidence indicates that citizens in Malawi respond to perceived economic conditions when forming judgments about government performance, reinforcing a growing body of research showing that voters in emerging democracies often evaluate governments based on perceived policy outcomes. Partisan identity operates as a stable and additive influence on satisfaction rather than as a conditional cognitive prior that varies with the clarity of economic signals. The study also highlights socio-demographic patterns: female respondents were significantly less likely than men to report satisfaction (odds ratio ≈ 0.75, p < 0.05), and lived poverty was negatively associated with satisfaction (odds ratio ≈ 0.84, p < 0.01), meaning individuals experiencing higher levels of deprivation were about 16 percent less likely to report satisfaction. The authors acknowledge limitations, including the use of partisan identification as a proxy for predictive expectations and the operationalization of ambiguity through neutral evaluations, and they call for panel surveys and experimental designs that directly measure expectations prior to evaluation. For now, the message from Malawi is clear: the economy beats the party, and ambiguity does not systematically amplify identity-based reasoning in this context—an important boundary condition in the relationship between partisanship and economic evaluations.

Subject of Research: Government's economic management and citizen satisfaction: a Partisan Disconfirmation Model

Article Title: Government's economic management and citizen satisfaction: a Partisan Disconfirmation Model

Article References: Banda, L. G., & Yakubu, K. (2026). Government's economic management and citizen satisfaction: a Partisan Disconfirmation Model. Frontiers in Social Psychology, 4, Article 1719133. https://doi.org/10.3389/frsps.2026.1719133

Image Credits: AI Generated

DOI: 10.3389/frsps.2026.1719133

Keywords: Government's, economic, management, citizen, satisfaction, Partisan, Disconfirmation, Model, scientific research