Impact Of Financial Influencers (Finfluencers) On Investment Decision-Making A Statistical Analysis

The democratisation of financial information through social media platforms has given rise to a new and influential category of content creators popularly known as financial influencers, or finfluencers. These individuals — ranging from certified financial planners and chartered accountants to self-taught retail investors and part-time market commentators — reach millions of retail investors daily through short-form videos, live streams, podcast episodes, Twitter threads, and Instagram posts covering a wide spectrum of topics including equity analysis, mutual fund reviews, cryptocurrency trends, and personal finance budgeting.

The rise of finfluencers must be understood against the backdrop of three major structural shifts in the financial information landscape. First, the exponential growth of mobile internet and social media penetration — particularly in emerging economies such as India — has created vast, previously underserved audiences hungry for accessible financial guidance. Second, the complexity of traditional financial advisory services has effectively excluded large swaths of the middle class and younger generations from personalised financial advice. Third, the COVID-19 pandemic catalysed an unprecedented retail investing boom globally, simultaneously driving demand for accessible investment content.

Despite growing academic and regulatory interest in the finfluencer phenomenon, empirical research using large-scale survey data and rigorous multivariate statistical methods remains limited. This study bridges that gap by administering a validated 13-item Likert-scale questionnaire to 348 participants and applying a comprehensive suite of inferential statistical tests. The primary objectives are: (i) to assess the descriptive profile of finfluencer engagement; (ii) to determine whether overall influence departs from a neutral baseline; (iii) to compare group means and categorical associations across demographics; (iv) to identify predictors of overall influence using multivariate regression; and (v) to present findings applicable to policy, practice, and future research.

METHODOLOGY

A cross-sectional survey research design was adopted. The structured questionnaire comprised 13 Likert-scale items (1=Strongly Disagree to 5=Strongly Agree) measuring finfluencer content consumption frequency, trust and credibility perceptions, influence on investment decisions, risk awareness, conceptual simplification, confidence building, and overall decision-making impact. The survey yielded 348 usable responses after data cleaning. The sample comprised predominantly young adults (94.3% aged 21–30), with 65% male and 35% female respondents. Students accounted for 76.1% of the sample.

Statistical Methods Employed

1. DESCRIPTIVE STATISTICS

Descriptive statistics provide the foundational quantitative summary of central tendency and variability for each Likert-scale item. Items related to the belief that finfluencers should be regulated (Mean = 3.20) and reliance on finfluencers over traditional advisors (Mean = 3.38) recorded the highest mean scores. The overall decision-making influence item recorded a mean of 2.681, below the neutral midpoint of 3.0. Skewness values for all items fall within the acceptable range of ±1.0, confirming that the normality assumption required for parametric tests is met.

Table 1: Descriptive Statistics of All Likert-Scale Items (N = 348)

Figure 1: Mean Scores of All 13 Likert-Scale Items with Error Bars (±1 SD)

Figure 2: Skewness and Kurtosis for All 13 Variables (Values within ±1 confirm normality)

2. Z-TEST: OVERALL FINFLUENCER INFLUENCE VS. NEUTRAL BENCHMARK

The one-sample Z-test determines whether the population mean of overall influence (x̄ = 2.681) differs significantly from the neutral midpoint (μ₀ = 3.0) of the five-point Likert scale. A mean significantly below 3.0 indicates that respondents perceive finfluencers as having a below-neutral — though not necessarily negligible — impact on their investment decisions.

1. Hypotheses

2. Results

3. Z-Test Hypothesis Curve

Figure 3: Normal Distribution Curve — Z = -5.37 falls far into the rejection region (shaded red)

4. Interpretation

The computed Z-statistic of -5.3728 substantially exceeds the critical value of ±1.96 for a two-tailed test at α = 0.05. The associated p-value of 0.0000 provides extremely strong statistical evidence against the null hypothesis. The evidence confirms that finfluencer content exerts a statistically significant overall influence on respondents' investment decision-making. The negative direction of the Z-statistic reflects that the observed mean falls below the neutral benchmark, indicating that while finfluencers are acknowledged as influential, their overall impact on decision-making is rated modestly below neutral. The effect size (Cohen's d ≈ 0.29) falls in the small-to-medium range, suggesting a meaningful but not transformative departure from neutrality — consistent with dual-process theories of persuasion that suggest media influence often operates through subtle shifts in attitude rather than dramatic behavioural change.

3. F-TEST (LEVENE'S TEST): EQUALITY OF VARIANCES BY GENDER

Levene's test for equality of variances assesses whether the variance in overall finfluencer influence scores is equal across male (n=226) and female (n=122) respondents. Equality of variance (homoscedasticity) is a prerequisite for independent-samples t-tests and validates parametric comparisons between gender groups.

1. Hypotheses

2. Results

3. F-Distribution Curve & Distribution Comparison

Figure 4: Left — Violin plot shows similar spread between genders; Right — F=0.915 lies well within acceptance region

4. Interpretation

Levene's F-statistic of 0.9154 with degrees of freedom (1, 346) yields a p-value of 0.5744. Since p > α = 0.05, we FAIL TO REJECT the null hypothesis. The variances in overall finfluencer influence scores are not significantly different between male and female respondents. Finfluencer content produces a similar degree of spread in both gender groups. This finding also validates the homoscedasticity assumption required for subsequent parametric comparisons between gender groups. One possible explanation is the homogenising effect of the shared social media environment — where both male and female respondents are exposed to similar algorithms, content formats, and influencer personas — which reduces gender-based differences in response variability.

4. CHI-SQUARE TEST: ASSOCIATION BETWEEN GENDER AND OCCUPATION

The Pearson Chi-square test of independence examines whether a statistically significant association exists between respondent gender (Male/Female) and occupational category (Others/Professionals/Staff/Student). A significant result would indicate that the occupational composition of finfluencer audiences varies systematically by gender.

1. Hypotheses

2. Observed Frequency Contingency Table

3. Chi-Square Test Results

4. Chi-Square Distribution Curve & Frequency Chart

Figure 5: Left — Observed frequencies by gender and occupation; Right — χ²=3.74 lies in acceptance region (df=3)

5. Interpretation

The Chi-square statistic of χ²(3) = 3.7448 yields a p-value of 0.2904. Since p > α = 0.05, we FAIL TO REJECT the null hypothesis. No statistically significant association was found between gender and occupational category. The effect size (Cramér's V = 0.104) is weak, confirming negligible practical magnitude. Both male and female respondents are similarly likely to be students, professionals, staff, or in other categories. This finding has an important practical implication: gender-based segmentation of financial communication interventions does not need to account for occupational composition differences, as finfluencer content appeals equally to male and female respondents regardless of their professional status.

5. ONE-WAY ANOVA: OCCUPATION GROUP DIFFERENCES IN OVERALL INFLUENCE

One-Way Analysis of Variance (ANOVA) tests whether the means of overall finfluencer influence scores differ significantly across four occupational groups: Others (n=13), Professionals (n=45), Staff (n=25), and Students (n=265). The F-statistic measures the ratio of between-group variance to within-group variance.

1. Hypotheses

2. Group Means by Occupation

3. ANOVA Summary Table

4. F-Distribution Curve & Group Means Chart

Figure 6: Left — Group means with SD error bars; Right — F=0.558 lies well within acceptance region (df₁=3, df₂=338)

5. Interpretation

The F-statistic of F(3, 338) = 0.5579 yields a p-value of 0.6431. Since p > α = 0.05, we FAIL TO REJECT the null hypothesis. Occupation does not significantly moderate finfluencer influence. The group means are closely clustered — Others (M=2.85), Staff (M=2.88), Students (M=2.68), Professionals (M=2.56) — suggesting that finfluencer influence is relatively occupation-agnostic. The slightly lower mean among professionals is consistent with the hypothesis that greater financial knowledge and professional experience attenuate finfluencer influence, though this difference is not statistically significant in this sample. Future research with more balanced occupational representation should use post-hoc Tukey HSD tests to explore specific pairwise differences.

6. TWO-WAY ANOVA: GENDER × INVESTMENT EXPERIENCE INTERACTION

Two-Way ANOVA simultaneously examines: (a) the main effect of Gender; (b) the main effect of Investment Experience (Below 1 Year / 1+ Years); and (c) the Gender × Investment Experience interaction effect on overall finfluencer influence. Interaction effects reveal whether the effect of one factor is moderated by another variable.

1. Hypotheses (Three Sets)

Hypothesis Set A: Main Effect of Gender

Hypothesis Set B: Main Effect of Investment Experience

Hypothesis Set C: Gender × Experience Interaction Effect

2. Results

3. Interaction Plot & F-Value Comparison

Figure 7: Left — Parallel interaction lines (non-crossing) confirm no significant interaction; Right — All F-values below critical threshold

4. Interpretation

None of the three effects tested reach statistical significance at α = 0.05. The non-significant interaction effect (F = 0.4427, p = 0.5063) is particularly noteworthy from a theoretical standpoint — it implies that the absence of a gender effect on finfluencer influence is consistent across both low-experience and high-experience investor subgroups. These findings challenge the intuitive expectation that novice investors would be substantially more susceptible to finfluencer influence than experienced ones. The binary experience categorisation (below vs. above one year) is acknowledged as a crude proxy; more granular measurement of financial sophistication would likely reveal more nuanced relationships.

7. PEARSON CORRELATION ANALYSIS

Pearson product-moment correlation analysis assesses the strength and direction of linear relationships between pairs of continuous variables. By convention: |r| < 0.30 = Weak; 0.30 ≤ |r| < 0.50 = Moderate; |r| ≥ 0.50 = Strong. Correlation was computed for all 13 Likert-scale items to assess bivariate linear relationships and validate construct coherence.

1. Hypotheses

2. Key Correlation Results

3. Correlation Heatmap (13×13 Matrix)

Figure 8: Full 13×13 Pearson Correlation Matrix — Predominantly positive weak correlations confirm construct coherence

4. Interpretation

The correlation matrix reveals that inter-item correlations are generally weak to moderate, suggesting that finfluencer engagement constructs are positively related but represent somewhat distinct dimensions rather than a single undifferentiated construct. The absence of strong negative correlations across the entire matrix suggests internal consistency — respondents who engage more with finfluencer content tend to rate its various dimensions positively across the board. The overall positive valence in finfluencer perception is consistent with confirmation bias literature, which suggests that media audiences preferentially consume and favourably evaluate content that aligns with their pre-existing attitudes. The absence of strong multicollinearity (r > 0.80) between predictors validates the stability of subsequent regression estimates.

8. MULTIPLE LINEAR REGRESSION: PREDICTORS OF OVERALL FINFLUENCER INFLUENCE

Multiple linear regression (MLR) with OLS estimation examines the relationship between nine predictor variables and the overall finfluencer influence on financial decision-making. The regression equation is: Overall Influence = β₀ + β₁(Content) + β₂(Videos) + β₃(Paid) + β₄(Rely) + β₅(Credibility) + β₆(Confident) + β₇(LT Planning) + β₈(Simplify) + β₉(RiskReturn) + ε

1. Hypotheses

2. Regression Model Summary

3. Coefficient Estimates

4. Regression Coefficient Chart

Figure 9: Ranked regression coefficients — Credibility (β=0.083) and LT Planning (β=0.076) are top positive predictors; Paid Promo Trust (β=-0.041) is the only negative predictor.

5. Interpretation

Despite the model's non-significance at the omnibus level, examination of individual coefficient patterns provides theoretically meaningful insights. The largest positive coefficients are associated with Credibility of Market Analysis (β=0.0826), Impact on Long-Term Planning (β=0.0755), and Simplify Complex Concepts (β=0.0547) — suggesting that affective trust, temporal orientation, and cognitive utility are the strongest individual pathways through which finfluencer content shapes overall investment decision-making. The negative coefficient for Paid Promotions Affect Trust (β=-0.0408) is the only negative predictor, confirming that concerns about commercial influences are associated with lower overall influence ratings. The very low R² (2.9%) warrants careful theoretical discussion: overall investment decision-making influence is a complex, multi-determined outcome shaped by variables not captured in this survey, including personality traits, financial literacy levels, social network effects, and market conditions.

1. MEGA TABLE: COMPREHENSIVE STATISTICAL FINDINGS SUMMARY

The following mega table synthesises all eight statistical analyses, presenting null and alternative hypotheses, key test statistics, p-values, decisions, effect sizes, and interpretive conclusions in a single integrated reference. This table serves as the definitive summary of the study's statistical findings.

Mega Summary Interpretation

The eight statistical analyses collectively paint a nuanced portrait: (1) Finfluencers DO exert a statistically significant, though below-neutral, influence on investment decision-making. (2) This influence is DEMOGRAPHICALLY UNIVERSAL — neither gender, occupation, nor investment experience significantly moderates finfluencer impact. (3) The primary psychological MECHANISMS of influence are affective trust (credibility) and cognitive utility (simplification + long-term orientation). (4) Commercial concerns about paid promotions DAMPEN influence, supporting mandatory disclosure policies. (5) The construct is COHERENT — all items are positively related, confirming the survey instrument's validity.

Mega Summary Charts

Figure 10: Mega Summary — (Top-Left) p-values for all 8 tests; (Top-Right) 1 significant vs. 7 non-significant tests; (Bottom-Left) Demographic mean comparisons; (Bottom-Right) Top regression predictors ranked

10. COMPREHENSIVE DISCUSSION & POLICY IMPLICATIONS

1. The Paradox of Significant But Below-Neutral Influence

The Z-test paradox: finfluencer influence is statistically highly significant (Z = -5.37, p < 0.001), yet the direction of that significance is below the neutral midpoint rather than above it. Respondents are not neutral about finfluencers — they have formed a clear opinion — but their overall assessment falls below the midpoint. This is consistent with the 'attitude-behaviour gap' in behavioural economics, and may also reflect social desirability bias. Longitudinal studies with objective portfolio tracking data would be needed to definitively resolve this interpretation.

2. The Universality of Finfluencer Appeal

A consistent pattern across F-test, Chi-square, One-Way ANOVA, and Two-Way ANOVA is the absence of statistically significant demographic moderation of finfluencer influence. Neither gender, occupation, nor investment experience significantly differentiates the extent or variability of finfluencer influence. This convergent pattern of null findings across four independent statistical tests is itself a substantive finding, consistent with the homogenising character of algorithm-curated social media environments. The policy implication is significant: protective measures must be designed for broad deployment rather than targeted application.

3. The Primacy of Affective Trust and Cognitive Utility

The regression analysis identifies perceived credibility (β=0.083), long-term planning impact (β=0.076), and conceptual simplification (β=0.055) as the strongest individual predictors. These map onto two psychological dimensions: (i) affective trust — the degree to which respondents trust the expertise and integrity of finfluencers; and (ii) cognitive utility — the perceived educational value in making complex financial concepts accessible. Finfluencers derive their influence not from institutional authority but from accessible explanations and long-term planning orientation that traditional financial advisors often struggle to provide.

4. The Commercial Trust Problem

The negative regression coefficient for Paid Promotions Affect Trust (β=-0.041) reveals that concerns about commercial underpinnings are negatively associated with overall influence ratings. While relatively small in magnitude, its direction is theoretically important: trust erosion through perceived commercial conflicts of interest does operate as a dampener of finfluencer influence. This provides empirical support for mandatory conflict-of-interest disclosure and advertising labelling requirements, though disclosure alone will not substantially neutralise finfluencer influence given the small effect size.

5. Policy Recommendations