R语言ltm包运行IRT模型报错：数据类型不符问题如何解决？

Hey there! Let's work through this frustrating error with the ltm package—this is a super common pitfall for new R users, so you're not alone. The error message says your data needs to be a numeric matrix/data frame with at least two columns, but you've already checked the column count and converted to a matrix. Let's break down the most likely fixes:

1. Verify Your Data Is Actually Numeric #

Even if your data looks like numbers, it might be stored as factor or character type (super easy to happen when importing data from CSVs). The ltm() function strictly requires numeric values.

• First, check your data structure with:

  str(your_data)
  

  Look for entries like Factor or chr instead of num.

• If you see non-numeric types, convert all columns to numeric. For a data frame, use:

  # Convert factors/characters to numeric
  your_data[] <- lapply(your_data, function(x) {
    if (is.factor(x)) as.numeric(as.character(x))
    else if (is.character(x)) as.numeric(x)
    else x
  })
  

  Note: If you have character values that can't be converted (like "N/A" strings), this will turn them into NA—that's okay, ltm can handle missing values if you set na.action = na.exclude in your function call.

2. Check for Hidden Non-Numeric Values #

Sometimes datasets have placeholder values like "?", "Missing", or even string "NA" instead of actual NA values. These will break the numeric requirement even if most values look like numbers.

• Scan for unusual values with:

  # Check unique values in each column
  lapply(your_data, unique)
  

• Replace any non-numeric placeholders with NA:

  # Example: Replace "?" with NA
  your_data[your_data == "?"] <- NA
  

3. Remove Columns with All Missing Values #

If one or more of your columns are completely filled with NA, ltm() will automatically drop them. This could leave you with fewer than two valid columns, triggering the error.

• Check for all-NA columns:

  colSums(is.na(your_data))
  

• Delete columns that have no valid data:

  # Keep only columns with at least one non-NA value
  your_data <- your_data[, colSums(is.na(your_data)) < nrow(your_data)]
  

4. Double-Check Your ltm() Syntax #

Make sure you're passing the entire dataset to the function correctly. It's easy to accidentally pass a single column instead of the full data frame/matrix.

• Correct syntax (for a 1-parameter logistic model):

  library(ltm)
  model <- ltm(your_data ~ z1, na.action = na.exclude)
  

• Wrong syntax (accidentally passing one column):

  # This will throw the error!
  model <- ltm(your_data$item1 ~ z1)
  

After trying these steps, run str(your_data) again to confirm you have a numeric data frame/matrix with at least two valid columns, then re-run your ltm() call.

内容的提问来源于stack exchange，提问作者Mpil