# Measure selection bias in accelerator programs. AI Agent Connect

> Accelerator Selection Bias Correction provides analytical tools to identify and correct selection bias in startup accelerator programs. Program managers often struggle to tell if a company's success is due to the program itself or if the company was already exceptional. This MCP helps you separate the two. By comparing the performance metrics of accepted companies against rejected applicants, you get a clear picture of your selection process's true value. You can find the statistical gap between cohorts, isolate the program's actual impact, and measure how well your current selection criteria predict future success. It's essential data for any program director.

## Overview
- **Category:** business-intelligence
- **Price:** Free
- **Endpoint:** https://edge.vinkius.com/vk_preview_J6hTHey2s2hm5A5AD3VAxmOAU5gxJa4FiXMPKEJD/ai-agent-connect
- **Tags:** accelerator, selection-bias, data-analysis, startup-metrics, predictive-modeling

## Description

When you run an accelerator, you want to know if the program is actually making companies better, or if you just picked the best ones to begin with. Selection bias is the biggest hurdle here. This MCP gives you the data to tackle that problem head-on. You connect your preferred AI client and ask it to analyze your cohort data. The tools let you move past simple success rates. Instead, you get quantifiable metrics that show the program's unique contribution. You can determine the true value-add of your curriculum and networking opportunities, separating that from the inherent excellence of the founders. This is critical for proving ROI and making data-driven changes to your intake process.

## Tools

### calculate_bias_magnitude
Calculate the magnitude of selection bias

### estimate_true_value_add
Estimate the true value-add of the accelerator

### evaluate_selection_effectiveness
Evaluate how well selection criteria predict success

## Prompt Examples

**Prompt:** 
```
How much selection bias is present in our current cohort?
```

**Response:** 
```
The selection bias magnitude is 0.15, indicating a significant gap between the accepted and rejected groups.
```

**Prompt:** 
```
What was the true value-add of the accelerator after adjusting for bias?
```

**Response:** 
```
The estimated true value-add is 25% uplift over the baseline success rate.
```

**Prompt:** 
```
How accurate are our selection criteria at predicting success?
```

**Response:** 
```
The selection model accuracy is 0.72, suggesting a strong predictive power for future outcomes.
```

## Capabilities

### Measure bias gap
The AI uses this MCP to calculate the statistical gap between accepted and rejected cohorts.

### Isolate program value
You can ask the AI to estimate the true value-add, removing the effect of pre-existing company excellence.

### Assess criteria power
The AI evaluates how well your current selection criteria predict a company's future success.

## Use Cases

### Program Review
After a cohort finishes, run the MCP to determine if the program's value-add justifies the operational cost.

### Intake Process Improvement
Use the bias tools to identify if your current application criteria are too loose or too restrictive.

### Reporting to Stakeholders
Generate data showing the true, unbiased impact of the accelerator for investors and partners.

### Benchmarking
Compare your program's selection bias magnitude against industry averages to spot weaknesses.

## Benefits

- You quantify the statistical gap between accepted and rejected applicants.
- The MCP isolates the program's actual impact, separating it from inherent company quality.
- You measure how accurately your selection criteria predict future success metrics.

## How It Works

Connecting this MCP is simple. You connect your preferred AI client to the Vinkius catalog, and the tools become immediately available. You then prompt your agent with specific questions about your cohort data.

1. Connect your AI client to the Vinkius catalog.
2. Reference the Accelerator Selection Bias Correction MCP.
3. Ask your agent to calculate the bias magnitude or true value-add.
4. The MCP runs the analysis and returns a specific, actionable metric.

## Frequently Asked Questions

**What is selection bias in this context?**
Selection bias means that the companies accepted into the program might already be better than the companies rejected. This MCP helps you measure that gap so you know the program's real contribution.

**Does this MCP tell me if my program is good?**
It gives you metrics to prove it. You can get estimates of the true value-add and see how well your selection criteria predict success. It's data, not a grade.

**Do I need to provide raw data?**
Yes. The tools require data comparing accepted and rejected applicants to calculate the bias magnitude and other metrics.

**What kind of data does it analyze?**
It analyzes performance metrics and success rates across different cohorts to determine statistical gaps and predictive power.

**Can I use this for non-accelerator programs?**
The tools are designed for selection bias correction, so they work whenever you need to compare two distinct groups based on a selection process.