UnboundLocalError: local variable referenced before assignment -- add as global variable?

Question

I am running Python v3.8.6. When I run the following code, it works.

def plot_timeseries(vars_to_plot, plt_config, plt_type=None):
    if plt_config['ggplot']:
        # Use ggplot style (grey background, white grid lines)
        plt.style.use('ggplot')
    # Use the first object in the list to parse common attributes.
    years = calc_year_span(vars_to_plot[0].start_year, vars_to_plot[0].end_year)
    units = vars_to_plot[0].units
    var_short = vars_to_plot[0].short_name
    var_long = getattr(Variables, var_short)['cesm_long_name']
    # Iterate over the variable objects & plot.
    if plt_type == 'diff':
        # Model config difference plot.
        diff_groups = group_vars_diff(vars_to_plot)
        for idx, (dataset, var_dict) in enumerate(diff_groups.items()):
            # Diff = perturbation - reference
            diff_cube = get_diff(var_dict['pert'], var_dict['ref'])
            plt.plot(years, diff_cube.data, linestyle=PlotStyle.styles[idx],
                     color=PlotStyle.colors[idx], label=dataset)
            if plt_config['write_data']:
                # Write var arrays to csv
                plt_config['config_alias'] = dataset
                save_plot_data(var_short, years, diff_cube.data, plt_config, plt_type='diff')
        default_plt_name = get_default_plot_name(var_short, plt_config, plt_type='diff')
    else:
        # Variable timeseries plot.
        for idx, var_obj in enumerate(vars_to_plot):
            plt.plot(years, var_obj.cube.data, linestyle=PlotStyle.styles[idx],
                     color=PlotStyle.colors[idx], label=var_obj.alias)
            if plt_config['write_data']:
                # Write var arrays to csv
                plt_config['config_alias'] = var_obj.alias
                save_plot_data(var_short, years, var_obj.cube.data, plt_config)
        default_plt_name = get_default_plot_name(var_short, plt_config)
    plt.xlabel('Year')
    plt.ylabel('{} ({})'.format(var_short, units))
    plt.title(plt_config['title'].format(var_obj.exp_full, var_long))
    plt.tight_layout()
    if 'ggplot' in plt_config and not plt_config['ggplot']:
        # Only call when not using ggplot style, otherwise no grid lines will be visible.
        plt.grid()
    plt.legend()
    # Adjust figure size
    fig = plt.gcf()
    fig.set_size_inches(12, 8)
    plt.close()

However, when I change one of the function parameters to plt_type='diff' and run it, I get this error: UnboundLocalError: local variable 'var_obj' referenced before assignment

The error log points to this line of the code: plt.title(plt_config['title'].format(var_obj.exp_full, var_long))

To remedy this error, I tried adding global var_obj within the function, but then I get this error: NameError: name 'var_obj' is not defined

What am I missing?